Journal Menu
`Advances in Mechanical EngineeringVolume 2013 (2013), Article ID 928671, 9 pageshttp://dx.doi.org/10.1155/2013/928671`
Research Article

## Mathematical and Simulation Modelling of Moisture Diffusion Mechanism during Plastic IC Packages Disassembly

Department of Mechanical Engineering, Tsinghua University, Beijing 100084, China

Received 31 May 2013; Revised 27 September 2013; Accepted 27 September 2013

Academic Editor: Feng Jiang

Copyright © 2013 Peng Mou et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

#### Abstract

Reuse of plastic IC packages disassembled from printed circuit boards (PCBs) has significant environmental benefits and economic value. The interface delamination caused by moisture diffusion is the main failure mode of IC packages during the disassembling process, which greatly reduces the reusability and reliability of disassembled IC packages. Exploring moisture diffusion mechanism is a prerequisite to optimize prebaking processes before disassembling that is an effective way to avoid the interface delamination. To this end, a computational model with variable boundary conditions is developed based on the different combination state of water in IC packages. The distribution characteristics and mechanism of moisture diffusion behavior are analyzed including the humidity distribution field and the relation among baking temperature, water loss rate, and baking time during baking process, and then the results are validated by FEA simulation based on the improved definition of relative moisture concentration. Baking under variable temperature is proposed and compared with the baking process and baking efficiency under constant temperature to find out the optimized baking parameters. Finally, a set of curves which indicate the relation between baking energy consumption and temperature are determined under actual industrial baking experiments, which could be used as references to develop industrial standards for PCB disassembling process.

#### 1. Introduction

Printed circuit board is the typical and fundamental component for almost all electronic products. With the rapid development and updating of electrical and electronic products, the amount of waste PCBs is increasing sharply and has caused severe environmental problems. Recycling and reusing waste PCBs have become a significant challenge in many countries. According to the statistical data, the average designed lifespan of plastic IC packages mounted on PCBs is about 500,000 hours, but their average service life is just around 20,000 hours, only 5% of their designed lifespan. Reuse of these plastic IC packages disassembled from PCBs will bring great environmental benefits and economic value.

Currently, polymer materials have been widely used in microelectronic IC packaging. Since the polymer materials are a kind of high porous material, the IC packaging is susceptible to moisture absorption and the moisture will condense in free volumes or nanopores in polymer materials and along interfaces [1]. The moisture concentration will reach to saturated state after serving for a long time under ambient environment. The condensed moisture will vaporize and diffuse during the traditional heating disassembling process, which makes the interfacial adhesion strength drops substantially, and delamination may occur at weak interfaces. Figure 1 shows the comparison of delaminated IC packages with normal IC packages by using ultrasound scanned pictures of a type of PQFP package.

Figure 1: Ultrasound scanned pictures of IC packages. (a) IC packages with delamination. (b) Normal IC packages.

When delamination occurs, the cracks caused by the interface delamination provided convenient intrusive channel for humidity and contaminative ions, which may accelerate the IC packages’ corrosion rate and reduce its lifespan and reliability. Moisture diffusion has great influence on the reliability of IC packages and is the main reason to induce interface delamination. In order to improve the reliability of IC package, it is necessary to study the mechanism of moisture diffusion induced interface delamination. In the past 10 years, many researches have been made on moisture diffusion mechanism and behavior for packaging reflow process.

Zhang et al. [1] integrated the thermomechanical and hygromechanical stress together to consider the non-uniform moisture distribution during reflow. The shear stress was found to be dominated along all the interfaces, and the molding compound/lead-frame interface around the junction of die attach fillet would be the initiation of delamination. Xie et al. [2, 3] developed direct concentration approach (DCA) to solve moisture diffusion problems under varying ambient temperature and humidity loading conditions. In DCA, the moisture concentration was used as a basic field variable, and constraint equations were applied at interfaces to satisfy the interface continuity requirement. DCA was then applied to 3D ultrathin stacked-die chip scale packages, and the results showed that a small reduction in substrate thickness could result in a significant decrease in moisture concentration. Zhao et al. [4] analyzed the effects of temperature and humidity on the failure of QFN by using moisture sensitivity analysis and wet-thermal simulation method. Mei and Yao [5] presented a combined numerical methodology to predict the thermal-humidity behavior of the Chip on Glass packaging process. Among these researchers, the research team coming from Lamar University has conducted many fruitful research works on moisture diffusion and interfacial delamination [68]. Fan et al. presented a micromechanics-based vapor pressure model to study the moisture impacts on package reliability. The model could describe the evolution of vapor pressure in voids. Effective stress concept was introduced to consider the effect of vapor pressure in the development of a continuum mechanics framework. Vapor pressure was considered as internal field variable, which is related to moisture evaporation. The latest research of FAN concentrated on the effects of temperature gradient on moisture diffusion in high power devices. Results showed that devices at ON condition operate at less humid ambient condition than at OFF condition because of the created local hot spots.

Although current researches on moisture diffusion are of high reference value, most of them focused on the surface mount reflowing process, which is different from actual dissembling process. As for the used IC packages, since the moisture concentration has reached to the saturated state and the long time working will cause some microcracks to gather inside IC packages, which make used IC packages more sensitive to moisture diffusion when compared with new IC packages. In order to improve the reusability and reliability of disassembled IC packages, it is very important to reduce the interface delamination during the disassembling process. Our previous experiments indicated that prebaking process before disassembling could effectively reduce the interface delamination failure [911].

In this paper, a computational model with variable boundary conditions is developed based on the different combination state of water in IC packages, which is used to study the distribution characteristics and mechanism of moisture diffusion behavior during prebaking process. Furthermore, the humidity distribution field and the relation among baking temperature, water loss rate, and baking time during baking process are analyzed based on the computational model, and the results are validated by FEA simulation and field experiments under variable temperature baking process. The optimized baking parameters are suggested based on FEA analysis and baking experiments.

#### 2. Theoretical Model of Moisture Diffusion

##### 2.1. Microscopic Mechanism of Moisture Diffusion

The packaging materials of the most IC packages are porous polymeric compound material. It is estimated that the density of water vapor absorbed in the polymer is 0.0802 mg/mm3, which is about 8% of the liquid water and 100 times of the standard water vapor [12], and the water molecules or moisture inside the polymeric compound is condensed into the mixed liquid/vapor two-phase state. When absorbed, the moisture is stored in micropores or free volumes in the polymer material. Some water molecules stayed freely inside the micropores of the compound materials which are called as unbounded state, and some other water molecules had connected with the hydrogen bond of NH4 and OH of the polymer compound which is called bounded state. When combined with the hydrogen bond of the compound, the absorbed water molecules will weaken the combination of compound macromolecule with SIO4 layer on the chip surface and then induce the interface delamination.

The main microfactors which affect the moisture absorption are polarity of the polymer material and the volume of the micropores. The diffusion coefficient of the nonamines compound materials is much larger than the amine-containing compound materials around 5°C. The reason for their difference on diffusion coefficient is that amine-containing compound materials provided more capture points which hindered the movement of the water molecules.

In fact, the moisture diffusion is not sensitive to the overall molecular thermal motion but restricted to the partial intense molecule thermal motion. The density and distribution of the gathered moisture stayed in the micropores of different material interface play a significant influence on the compound properties [12].

##### 2.2. Mathematical Definition of Moisture Diffusion

The moisture diffusion is the mass transfer process of water molecules, and it met FICK’s Second Theorem shown in where is the diffusion coefficient (mm2/s) and is humidity (mg/mm3). changes with temperatures, and it is decided by the properties of the chip material and meets Arrhenius Equation shown in where is the gas constant (8.31441 J/K·mol), is the Kelvin degree, and is the activation energy of diffusion coefficient.

From (2), is a function of . Usually, used IC packages are baked under constant temperature and is a constant value. For most IC packages, the length is about 25–30 mm and the width is about 14–16 mm, which is much larger than the thickness, about 2-3 mm, so the heterogeneity on length and width direction could be ignored and the diffusion process could be simplified only along the thickness direction. Then, the equation could be simplified as The upper and lower surfaces of the IC chips are exposed to the same humidity environment and the moisture diffusion on both sides has symmetrical features, so using half of the chip is enough to calculate and analyze the diffusion process. The driving force of moisture diffusion at the boundary is the difference between real-time concentration and the final equilibrium concentration.

The distribution function of moisture humidity along the thickness direction could be described as at the initial state, and the boundary conditions and initial conditions of moisture diffusion in IC packages are shown in where is the diffusion speed (mm/s). Equations (3)-(4) are general description of moisture diffusion and could be used to solve distribution of moisture and humidity at specific time.

##### 2.3. Mathematical Solving of Moisture Diffusion

The commonly used PQFP IC packages ITE8705 is selected as the studying object, whose appearance and structural sectional view are shown in Figure 2.

Figure 2: Appearance and sectional view of ITE8705. (a) Appearance of ITE8705. (b) Sectional view of ITE8705.

According to the actual baking experiments, the moisture content curve has great differences under different baking process and IC packages. The differences may be caused by the variable boundary condition. It is not possible for the moisture content at the boundary to mutate when external conditions change. There should be continuous changing transitional boundary layer around the IC packages. The properties of boundary layer, including press, ambient humidity, temperature, and air flow rate during the baking process, have great impact on diffusion speed (). At the initial moment, the humidity has an initial value and then would diffuse continuously until reaching to an equilibrium concentration .

As for moisture diffusion, moisture concentration is the driving forces of transmission process, which are decided by the difference of concentration inside IC packages. Since larger difference will induce greater driving force, the moisture diffuses at very high speed at the initial stage. The diffusing speed will slow down since the concentration is reaching to an equilibrium state gradually. So, the moisture concentration decreased very fast at the initial stage and the speed of decreasing will slow down gradually, which is monotonically decreasing function.

The vapor pressure is directly affected by the moisture concentration and disassembling temperature. Moisture concentration will affect the vapor pressure during the disassembling process at higher temperature and then decides whether the interface delamination would arise inside the IC packages. The moisture concentration gradient from the middle to surface of IC packages after baking for 0.5 h at 125°C is shown in Figure 3.

Figure 3: Distribution of moisture concentration.

Figure 3 shows that moisture concentration is very high at the inside middle layer, and it decreased very quickly from internal to external surface. Since bonding layer is in the central part of the IC packages, it has small connection with the plastic module, which causes very large diffusion resistance. So, moisture concentration is very high at the inside middle layer and it may form greater local vapor pressure, which may cause potential microcracks or even interface delamination. Since the used IC packages are more sensitive to the vapor pressure, it is better to reduce the baking temperature to ensure its reliability.

In fact, it is very difficult to make real-time measuring of the moisture concentration, but the moisture content could be obtained directly by weighing during the baking process. It is a long time to bake the IC packages to totally dried state. Even if the IC packages are baked under 125°C, it still needs about 120 hours to dry it and the time and economic costs are too high to be acceptable.

The relation among baking temperature, baking time, and baking effect is shown in Figure 4.

Figure 4: Baking temperature and time curve clusters.

Figure 4 shows that baking time is very sensitive with baking temperature below 100°C. The reducing rate of baking time has very little difference when baking temperature is higher than 150°C. So, it is not advisable to reduce baking time by increasing baking temperature above 150°C.

#### 3. FEA of Moisture Diffusion

##### 3.1. Basis of FEA Simulation

The moisture diffusion process has similar control equation with heat transfer process according to (1), and it is possible to use the thermal analysis module of FEA software to analyze the moisture diffusion process.

The comparison of the heat transfer and moisture diffusion is listed in Table 1.

Table 1: Comparison of heat transfer and moisture diffusion [13].
##### 3.2. Improved Definition of Relative Concentration

For IC packages, the moulding compound and bonding layer are the main moisture absorber. The moisture concentration at their boundary is not continuous because of the differences of their physical properties.

The significant difference between heat transfer and moisture diffusion process is the continuity of the field variables, and . The temperature is continuous during the heat transfer process, but the moisture concentration is discontinuous at the interface of different materials, as shown in Figure 5, and made it inadmissible to FEA treatment.

Figure 5: Discontinuous moisture concentration at different material interface [14].

In order to solve this problem, Wong proposed the definition of relative concentration (wetness) as follows [15]: If is continuous in different medium ( and ), then the hypothesis could be described as The above hypothesis could be proved by Henry Partial Pressure Theorem. According to the actual disassembling and baking process, when moved into the drying cabinet, the saturated moisture concentration of IC packages is equal to the outside water vapor concentration . The improved definition of the initial relative concentration is shown in and the relative concentration on the boundary is As shown in Figure 6, the moisture concentration of the IC packages would equal the vapor concentration in the external environment during a long time of diffusion, and the final relative concentration would be , instead of .

Figure 6: Continuous relative concentration at different medium inside IC packages.

When substituting relative concentration () for absolute humidity (), (1) could be converted into a relative concentration diffusion differential equations, which is identical with heat transfer process, and the thermal analysis module could be used to analyze the moisture diffusion process.

##### 3.3. Simulation of Moisture Diffusion at Constant Temperature

The studying object, IC packages ITE8705 (shown in Figure 1), has symmetrical shape, and the gradient direction of moisture diffusion mainly concentrates on the thickness direction. When building the FEA model, half of the IC package section is enough for calculation and analysis.

Currently, there are no available baking standards for PCB disassembly process, but the industry standard (IPC/JEDEC J-STD-033A) for baking new IC packages could be used for reference. Since the disassembled IC packages are very sensitive to the moisture due to the potential existence of internal microcracks, the baking temperature should be carefully selected when considering the efficiency and reliability. In this study, 90°C and 125°C are selected for moisture diffusion analysis from the four temperature levels defined in IPC/JEDEC J-STD-033A (40°C, 90°C, 125°C, and 150°C). The assumptions and experimental conditions are as follows.(i)Assuming that disassembled IC packages have worked or have been stored at 25°C/60% RH for a long time, they have reached saturated moisture concentration.(ii)The temperature of baking ovens is set to 90°C/1.5% RH and 125°C/1.5% RH separately.

Humidity distribution field of IC packages after baking at 90°C for 120 hours is shown in Figure 7.

Figure 7: Humidity distribution field of IC packages after baking at 90°C for 120 hours.

Bonding layer is in the central part of the IC packages and it has small connection with the plastic module, which caused very large diffusion resistance. Though the moisture diffusion coefficient of the bonding layer is greater than the plastic module, the speed of its moisture diffusion is the slowest due to the restriction of its geometry and location. So, the bonding layer area has the highest moisture concentration. It needs about 288 hours to bake the IC packages to dry state at 90°C. Humidity distribution field of IC packages after baking at 125°C for 120 hours is shown in Figure 8.

Figure 8: Humidity distribution field of IC packages after baking at 125°C for 120 hours.

Generally, higher baking temperature will cause greater temperature gradient along the thickness direction of IC packages, which will increase the driving force of moisture diffusion and then accelerate the diffusion speed. Figure 8 indicates that the diffusion speed increased quickly with the temperature, and the balance time for baking has been shortened greatly.

##### 3.4. Simulation of Moisture Diffusion at Variable Temperature

According to the relation among baking temperature, time, and moisture concentration, the baking process at constant temperature has very complicated impact on the efficiency, energy, and baking time. The optimization of these indicators is contradictory. In order to resolve these contradictions, baking at variable temperature could achieve the integrated optimization.

The variable baking parameters are determined according to the following two rules.(i)Reduce the moisture changing rate at the initial stage to ensure the chip’s security and integrity.(ii)Accelerate the moisture changing rate at the second stage to reduce the balancing time for moisture diffusion.

Based on the above two rules, a group of parameters listed in Table 2 are selected for variable baking simulation.

Table 2: Time and temperature setting of variable baking.

Similarly, the assumptions and experimental conditions are the same as baking at constant temperature. Because the balancing humidity is different under 90°C and 125°C and the change of relative concentration is not continuous, the final results are expressed by absolute humidity. Taking the 7th parameter as the example, Figure 9 indicates the baking simulation process under 90°C (6 h) and 125°C (66 h).

Figure 9: Humidity distribution field of IC packages after baking at variable temperature under 90°C (6 h)–125°C (66 h).

In the first 6 hours, moisture diffusion process is consistent with the baking process at constant temperature. When baking temperature rises from 90°C to 125°C at the second stage, higher baking temperature will cause greater temperature gradient along the thickness direction of IC packages, which will increase the driving force of moisture diffusion. So, the increase of the temperature accelerates the moisture diffusion speed and moisture changing rate sharply. After baking for 36 hours, the humidity is at very low level and the increasing of temperature has very limited impact on the diffusion speed of moisture.

In order to secure the reusability and reliability of the IC packages, it is rational to bake them under 90°C for 7 to 10 hours. In the second baking stage, baking temperature could be increased around 125°C to accelerate the moisture diffusion speed and shorten the baking time for economic and cost reasons.

#### 4. Results and Discussion

According to the above FEA simulation, baking at variable temperature has greater advantage than baking at constant temperature when considering the baking time and the IC package’s integrity and reliability. Economic and cost aspects should be carefully evaluated before actual industrial application. Baking process consumed a lot of energy which should be optimized to make it affordable.

The baking power increases with baking temperature during the baking process and then decreases the baking time. The relation between total baking energy and baking temperature is complicated, which is decided by the function relation among baking power, baking time, and baking temperature. In order to evaluate the energy consumption and baking cost, a commonly used industrial blast electric oven (DHG-9053A), shown in Figure 10, is selected for building energy consumption model.

Figure 10: Blast electric oven.

The energy consumption of the blast electric oven is mainly used to drive the blower and heating to maintain the temperature. The energy consumption model could be described in where is the thermal conductivity coefficient of the oven, is the conductivity area, is the thickness, is the air specific heat capacity, is the air mass exchanged within unit time, is the internal oven temperature and is the outside air temperature, and is the other energy consumed during the baking process.

According to (9), the energy consumption has a direct proportion with the temperature difference between internal oven and external environment. For a given oven, the characteristic parameters are determined and could be measured by baking experiment.

The baking time is decided by the baking results (usually expressed as water loss rate). For the selected IC package (ITE8705), the relation between water loss rate and baking time has been tested and listed in Table 3.

Table 3: Relation between water loss rate and baking time.

In fact, the difference of absolute water loss is not very significant at different waster loss rate. Since the baking time varies greatly from 90% to 99.9999%, it is necessary to define reasonable baking parameters to make it acceptable for industrial application.

In the premise of reducing interface delamination and improving the reliability of the IC packages, the relation among baking temperature, water loss rate, and baking time during baking process is confirmed by experiments. The temperature and power parameters are listed in Table 4.

Table 4: Temperature and average power baking process.

The relation between baking energy consumption and temperature under different water loss rate is tested and shown in Figure 11.

Figure 11: Relation between baking energy consumption and temperature under different water loss rate.

Figure 11 indicates that baking energy at different baking temperature has not been much different at the beginning stage, so it is not advisable to increase the temperature of the previous baking stage, and higher temperature will not reduce the energy consumption and may increase the possibility of interface delamination. The baking standards which are defined by the final moisture content had significant influence on the heating time as well as the consumed baking energy. The tested curve clusters could be used as references to develop industrial standards for PCB disassembling process when high IC packages reliability and reusability are required.

#### 5. Conclusions

This paper studies the moisture diffusion behavior during PCB disassembling process by building computational model with variable boundary conditions. The distribution of moisture concentration and relation among baking temperature, baking time, and baking effect have been confirmed based on the combination of the model and baking experiments. It is found that the resistance for moisture diffusion mainly came from the internal part of the IC packages. Then, improved definition of the relative concentration is defined and the model is validated and applied by FEA simulation. When compared with constant temperature, baking at variable temperature could achieve optimized efficiency and energy consumption. In order to secure the reusability and reliability of the IC packages, it is rational to bake them under 90°C for 7 to 10 hours, and then the baking temperature could be increased around 125°C to accelerate the moisture diffusion speed and shorten the baking time for economic and cost reasons. Finally, a set of curves which indicates the relation between baking energy consumption and temperature is determined under actual industrial baking experiments, which could be used as references to develop industrial standards for PCB disassembling process.

The impact of moisture humidity on crack propagation is still uncertain and the microscopic mechanism of their relation should be a research focus in the near future.

#### Nomenclature

 D: Diffusion coefficient (mm2/s) C: Moisture concentration (mg/mm3) R: Gas constant (8.31441 J/K·moL) H: Diffusion speed (mm/s).

#### Acknowledgments

The research is supported by the NSFC project (Grant no. 51075233) and National Key Technology R&D Program (Grant no. 2011BAF11B06).

#### References

1. M. Zhang, S. W. R. Lee, and X. Fan, “Stress analysis of hygrothermal delamination of Quad Flat No-lead (QFN) packages,” in Proceedings of the ASME International Mechanical Engineering Congress and Exposition (IMECE '08), pp. 339–347, Boston, Mass, USA, October-November 2008.
2. B. Xie, X. J. Fan, X. Q. Shi, and H. Ding, “Direct concentration approach of moisture diffusion and whole-field vapor pressure modeling for reflow process—part I: theory and numerical implementation,” Journal of Electronic Packaging, vol. 131, no. 3, Article ID 031010, 7 pages, 2009.
3. B. Xie, X. J. Fan, X. Q. Shi, and H. Ding, “Direct concentration approach of moisture diffusion and whole-field vapor pressure modeling for reflow process—part II: application to 3D ultrathin stacked-die chip scale packages,” Journal of Electronic Packaging, vol. 131, no. 3, Article ID 031011, 6 pages, 2009.
4. Y. Zhao, C. Chang, H. Chen, and G. Gao, “Simulation analysis of delamination failure and thermal-moisture,” Semiconductor Technology, vol. 35, no. 6, pp. 550–554, 2010.
5. Y. Mei and X. Yao, “A numerical method on thermal-humidity behavior of electronic packaging,” in Proceedings of the 12th International Conference on Electronic Packaging Technology and High Density Packaging (ICEPT-HDP '11), pp. 1044–1048, August 2011.
6. X. J. Fan, J. Zhou, G. Q. Zhang, and L. J. Ernst, “A micromechanics-based vapor pressure model in electronic packages,” Journal of Electronic Packaging, vol. 127, no. 3, pp. 262–267, 2005.
7. X. Fan and J.-H. Zhao, “Moisture diffusion and integrated stress analysis in encapsulated microelectronics devices,” in Proceedings of the 12th International Conference on Thermal, Mechanical and Multi-Physics Simulation and Experiments in Microelectronics and Microsystems (EuroSimE '11), April 2011.
8. X. Fan and C. Yuan, “Effect of temperature gradient on moisture diffusion in high power devices and the applications in LED packages,” in Proceedings of the Electronic Components & Technology Conference, pp. 1466–1470, Las Vegas, Nev, USA, May 2013.
9. X. Ding, X. Liu, D. Xiang, J. Yang, and G. Duan, “Research on delamination failure of plastic IC packages in components reuse processes,” in Proceedings of the 4th World Congress on Maintenance, Haikou, China, November 2008.
10. X. Ding, Research on delamination problem in plastic IC packages during PCB disassembly [M.S. thesis], Tsinghua University, 2009.
11. Z. Z. Liu, Moisture diffusion modeling and optimization of the baking process for disassembled IC packages [Bachelor thesis], Tsinghua University, 2011.
12. X. Cai, W. Huang, B. Xu, and Z. Cheng, “Morphology of the water in plastic electronic packaging materials,” Chinese Journal of Materials Research, vol. 16, no. 5, pp. 507–511, 2002.
13. E. H. Wong, S. W. Koh, K. H. Lee, and R. Rajoo, “Advanced moisture diffusion modeling & characterisation for electronic packaging,” in Proceedings of the 52nd Electronic Components and Technology Conference, pp. 1297–1303, May 2002.
14. J. Bie, X. Sun, and S. Jia, “Recent advances in the study of moisture absorption in plastic electronic packaging,” Advances in Mechanics, vol. 37, no. 1, pp. 35–47, 2007.
15. E. H. Wong, S. W. Koh, K. H. Lee, K.-M. Lim, T. B. Lim, and Y.-W. Mai, “Advances in vapor pressure modeling for electronic packaging,” IEEE Transactions on Advanced Packaging, vol. 29, no. 4, pp. 751–759, 2006.